Brennan Center for Justice v. New York Police Department
On December 20, 2016, the Brennan Center filed an Article 78 action against the New York Police Department (NYPD) for failing to respond adequately to a Freedom of Information Law (FOIL) request for records.
The request, originally filed in June 2016, was for records regarding the acquisition and use of predictive policing technology. Predictive policing employs computer modeling to try to anticipate crime and to inform deployment of police resources. Algorithms analyze various datasets to predict where, what, and sometimes even who may be involved in a future crime, but there are significant concerns that they may also replicate existing patterns of discrimination.
Publicly available purchase orders showed that the City of New York had paid over $2.5 million to Palantir Technologies since 2012 for software, maintenance, and support. Palantir markets its technology as a platform designed to allow law enforcement to pull in data from a wide range of sources to assist with case management. Palantir also offers a predictive policing feature that suggests geographic areas where crimes are more likely to occur. The design and implementation of this technology could materially affect the civil rights of New Yorkers, but there is little public information about the actual costs and parameters of the NYPD’s implementation of predictive policing technology.
The original request was denied in its entirety in late June. The Brennan Center appealed the decision, and NYPD issued a blanket denial of the appeal on August 15, 2016. The Brennan Center sought to have the court compel the NYPD to disclose the documents required by FOIL and/or to comply with FOIL’s requirement that it describe in more detail that documents that it is not disclosing.
Read the Verified Petition.
Read the Memorandum of Law.
On April 7, 2017, the NYPD responded to the Brennan Center’s lawsuit and agreed to provide some documents in response to the Brennan Center’s request, including some vendor agreements and email communications with predictive policing software vendors (which were mostly redacted), a set of privacy protection guidelines from 2009, and a copy of an article on the department’s Domain Awareness System. They also disclosed that they do not use Palantir for this purpose, and instead use a predictive policing system that was developed in-house, after testing products from three different vendors. NYPD maintained, however, that they should not be required to produce any documents about their actual testing of commercial products or their current use of predictive policing software; they argued that information about tests of commercial products could reveal vendor’s trade secrets, while information about inputs and algorithms in use in the department would potentially allow criminals to replicate the results and predict police movements.
The Brennan Center does not believe that the information it is seeking would implicate vendor trade secrets or allow criminals to replicate the NYPD’s results. Nevertheless, it narrowed its request in its May 18, 2017 reply to focus on records reflecting testing and utilization of the predictive policing system, as well as the historical inputs and outputs of the model. Release of this information would not implicate current or future investigations without the underlying code, but could provide a basis from which to understand how the tools are generally being employed at the NYPD. The Brennan Center also renewed its request for audits of the system and up-to-date governing policies and procedures. Finally, the Brennan Center requested that the NYPD produce relevant vendor communications and redact trade secrets information, if any, rather than withholding them entirely. Disclosure of these communications would significantly contribute to transparency regarding the NYPD’s goals for and use of the software.
Read the Brennan Center’s Reply.
Read the Transcript of Oral Argument.
The Brennan Center is represented on this matter by Quinn Emanuel Urquhart & Sullivan, LLP.
Citizens deserve to know how their police department is allocating resources to fight crime. The Brennan Center will continue to fight for public release of information about technologies that implicate civil liberties.
On December 27, 2017, the New York State Supreme Court ordered the NYPD to produce certain records about the testing, development, and use of predictive analytics tools to the Brennan Center. Both parties filed cross motions for reargument, which were denied in July of 2018.
Read the December 27, 2017 Order here.
Read the Brennan Center’s Memorandum of Law in Support of Motion for Limited Reargument here.
Read the NYPD’s Affidavit in Support of Motion for Leave to Reargue here.
See the orders denying the cross motions for re-argument here.
Finally, after more than two years of litigation, the NYPD started releasing records detailing the department’s testing of three commercial predictive policing programs and its ultimate decision to build in-house the predictive software it would use to determine officer deployment. The records include internal communications and notes; promotional materials, email exchanges, and contracts from outside software vendors; and numerous output maps and data collected by the department’s in-house algorithms. In all, the Brennan Center received approximately 2,600 pages of information. This trove provides a crucial view into the mechanics of predictive policing, contributing to a more informed public discourse about the costs and consequences of such programs.
NYPD did not relinquish this information easily. Even after the December 27, 2017 court ruling required the department to release records satisfying the Brennan Center’s request, the department did not begin producing documents until November of the following year. And the NYPD did not finish releasing the records until April 2019, just short of three years after the Brennan Center filed its original FOIL request. The department’s obstructionist approach is concerning because citizens and watchdog organizations should not have to wait years for insight into the technologies shaping police behavior. Nevertheless, the documents represent a victory for members of the public who deserve to know how their police department is allocating its resources, and for civil liberties advocates committed to understanding the implications of predictive policing for individual freedoms.